How to find a refrigerant leak
Start by proving there is a leak
Before hunting a location, establish that the system genuinely does not hold pressure. Pressurise with dry nitrogen to an appropriate test pressure, isolate, and leave it with a gauge on.
Nitrogen is used rather than refrigerant for three reasons: it is inert and dry, it is far cheaper than losing a charge, and venting refrigerant to atmosphere for testing is exactly the practice the phase-down exists to stop. Correct for ambient temperature change over the test — a system standing in the sun will show a pressure rise that is thermal, not a sealed-system verdict.
A system that holds nitrogen overnight does not have a significant leak. If it drops, you now have a reason to spend the afternoon looking.
The methods, and what each is good for
| Method | Finds | Limitations |
|---|---|---|
| Standing pressure (nitrogen) | Whether a leak exists at all | Tells you nothing about where |
| Electronic detector | Small leaks, quickly, over a wide area | False positives near solvents and oils; needs periodic calibration; struggles in wind |
| Bubble solution | Pinpoint confirmation at a joint | Only works where you can reach and see; misses very small leaks |
| Ultrasonic | Leaks behind panels, in noisy plant rooms, at distance | Needs a pressure differential; background ultrasound interferes |
| UV dye | Intermittent and vibration-dependent leaks, over days or weeks | Slow; must be compatible with the oil; leaves residue |
| Oil traces | Long-standing leaks — oil escapes with refrigerant | Only after the leak has run a while |
The practical sequence most experienced technicians settle on: nitrogen test to confirm, electronic detector to narrow down the area, bubble solution to pinpoint the joint. Dye when the leak is intermittent and the first pass found nothing.
Where leaks actually are
Leaks are not randomly distributed. They cluster where the system has been worked on and where it moves.
- Flare joints — the single most common site, especially where the flare was made badly or over-torqued
- Brazed joints, particularly ones made in awkward positions on site rather than in a workshop
- Schrader valves and cores — frequently the leak, and frequently overlooked because they look trivial
- Shaft seals on open-drive compressors
- Anywhere subject to vibration — which is why transport units and poorly-mounted outdoor units lead the statistics
- Coil corrosion, especially in coastal or chemically aggressive atmospheres
The pattern to notice is that most of them are consequences of installation quality. A system assembled carefully leaks far less than an identical one assembled in a hurry, and no amount of detection equipment retrofits that.
Record it, because it is an emission
Every leak is a greenhouse-gas release, and for many organisations a reportable one. Fugitive refrigerant is a Scope 1 emission, and the quantity you recharged is the best evidence of the quantity that escaped.
What to write down at every recharge: the date, the system, the refrigerant, the quantity added in kilograms, and what was found and repaired. That record does three jobs at once — it feeds the emissions inventory, it tells you which machines are getting worse, and it stops the annual top-up becoming invisible routine. Turning leaked kilograms into tonnes of CO2e.
The commercial version of the same point: if you are buying the same gas for the same machine every season, the gas is the cheapest part of what that machine is costing you.
Quick answers
What is the best way to find a refrigerant leak?
There is no single best method. Confirm the leak exists with a dry-nitrogen standing-pressure test, narrow the area with an electronic detector, then pinpoint the joint with bubble solution. Use UV dye for intermittent leaks that the first pass cannot find.
Why use nitrogen rather than refrigerant to pressure-test?
Nitrogen is inert, dry and inexpensive, and testing with refrigerant means venting it to atmosphere, which wastes gas and releases a greenhouse gas unnecessarily.
Where do refrigerant leaks usually occur?
Most commonly at flare joints, site-made brazed joints, Schrader valve cores, compressor shaft seals, and anywhere subject to vibration. Coil corrosion is a frequent cause in coastal or chemically aggressive environments.
- ASHRAE · leak detection and refrigerant-handling practice · ashrae.org.
- US EPA · Section 608 leak-repair trigger rates and leak-rate calculation methodology · epa.gov.
- GHG Protocol · accounting for fugitive refrigerant emissions · ghgprotocol.org.
This guide is general information compiled from the cited sources, not legal, safety or engineering advice. Read our full disclaimer.
Related: Refrigerant leaks and Scope 1 · Evacuation and moisture · Cylinder storage safety · Tools and accessories · CO2e calculator